instance.host not cleared after reschedule

[Joe Cropper]

For example, take this scenario:

1. Assume an environment with a single host… call it host-1
2. Deploy a VM, but force an exception in the spawn path somewhere to simulate some "hypervisor error”
3. The scheduler correctly attempts to reschedule the VM, and ultimately ends up (correctly) with a NoValidHost error because there was only 1 host
4. However, the instance.host (e.g., [nova show <vm>]) is still showing ‘host-1’

After discussion on the mailing list and IRC, we decided this should be considered a bug since the instance is no longer tied to the host since it blew up in spawn and tried to reschedule. As such, we should clear it out when the claim is aborted.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/161069

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/161069
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=5d8eb7184ebcf0a1ad245de28d6405bda385b680
Submitter: Jenkins
Branch: master

commit 5d8eb7184ebcf0a1ad245de28d6405bda385b680
Author: Joe Cropper <email address hidden>
Date: Tue Mar 3 19:46:21 2015 -0600

    Nil out inst.host and inst.node when build fails

    This patch addresses a bug that leaves instance.host/node set after
    a VM fails to build or reschedule (e.g., due to an error in the spawn
    path). Rather, we should nil these out at the end of the build
    process so that the instance is not erroneously being associated
    with stale host/node information.

    Also included are a few cosmetic updates in comments and eliminating
    some implied 'None' parameters from "dict.get(key, None)" calls I
    found during the course of investigating this change.

    Change-Id: I4d9c90b1d3fcce5237323fffcc6db1fb93d23e37
    Closes-Bug: #1427944